Content Based Image Retrieval Using Local Derivative Patterns

A new image indexing and retrieval algorithm known as local derivative pattern (LDP_16_2) is proposed in this work. LDP_16_2 histograms are used as features of each image in the data base. LDP_16_2 encodes the higher order derivative information which contains more detailed discriminative features. This property made it a powerful tool for feature extraction of images in the data base. Improved results in terms of retrieval efficiency and computational complexity are observed over recent work based on LBP_16_2 (Local Binary Patterns) features based CBIR system and LBP correlogram features based CBIR system.. The distance measures viz. city block distance, Euclidean distance, Canberra distance and d1 2 distance are used as similarity measures in the proposed CBIR system. Superiority of d1 2 distance is observed over other distances in terms of average retrieval rate. .